// Policy iteration bug fixed in revision 7669

mdp

module strips
state : [ 0 .. 5 ] init 0;

[ takebox ] state=0 -> 1: (state'=1) ;
[ askforhelp ] state=0 -> 0.4: (state'=2)  + 0.6: (state'=0) ;
[ takeboxwithhelp ] state=2 -> 1: (state'=3) ;
[ takebox ] state=2 -> 1: (state'=3) ;
[ askforhelp ] state=2 -> 0.4: (state'=2)  + 0.6: (state'=2) ;
[ takebananas6 ] state=3 -> 0.3: (state'=4)  + 0.5: (state'=3)  + 0.2: (state'=1) ;
[ takeboxwithhelp ] state=3 -> 1: (state'=3) ;
[ takebox ] state=3 -> 1: (state'=3) ;
[ takebananas1 ] state=3 -> 0.1: (state'=4)  + 0.9: (state'=3) ;
[ askforhelp ] state=3 -> 0.4: (state'=3)  + 0.6: (state'=3) ;
[ takebox ] state=1 -> 1: (state'=1) ;
[ takebananas1 ] state=1 -> 0.1: (state'=5)  + 0.9: (state'=1) ;
[ askforhelp ] state=1 -> 0.4: (state'=3)  + 0.6: (state'=1) ;
endmodule

rewards
[ takebox ] true : 8;
[ askforhelp ] true : 5;
[ takeboxwithhelp ] true : 4;
[ takebananas6 ] true : 3;
[ takebananas1 ] true : 3;
endrewards